Top-down effects of predators can have important consequences for ecosystems. Insectivorous birds frequently have strong predation effects on herbivores and other arthropods, as well as indirect effects on herbivores' host plants. Diet studies have shown that birds in temperate ecosystems consume arthropods in winter as well as in summer, but experimental studies of bird predation effects have not attempted to quantitatively separate winter predation impacts from those in summer. To understand if winter foraging by insectivorous birds has consequences for arthropods or plants, we performed a meta-analysis of published bird exclusion studies in temperate forest and shrubland habitats. We categorized 85 studies from 41 publications by whether birds were excluded year-round or only in summer, and analyzed arthropod and plant response variables. We also performed a manipulative field experiment in which we used a factorial design to exclude birds from Quercus velutina Lam. saplings in winter and summer, and censused arthropods and herbivore damage in the following growing season. In the meta-analysis, birds had stronger negative effects on herbivores in studies that included winter exclusion, and this effect was not due to study duration. However, this greater predation effect did not translate to a greater impact on plant damage or growth. In the field experiment, winter exclusion did not influence herbivore abundance or their impacts on plants. We have shown that winter feeding by temperate insectivorous birds can have important consequences for insect herbivore populations, but the strength of these effects may vary considerably among ecosystems. A full understanding of the ecological roles of insectivorous birds will require explicit consideration of their foraging in the non-growing season, and we make recommendations for how future studies can address this.

1. Leaf shelter construction by herbivorous insects can improve leaf quality, sometimes changing resultant herbivory. In two experiments in a Missouri (USA) deciduous forest we quantified the impact of leaf tie construction and changes to leaf quality on subsequent leaf damage.2. First, using eight Quercus species, we compared damage to single leaves versus experimental leaf ties that had been stocked with either Pseudotelphusa quercinigracella (Gelechiidae) or Psilocorsis cryptolechiella (Depressariidae) to determine how initial leaf quality (total phenolics) influenced damage caused by shelter inhabitants.Skeletonization by leaf tying caterpillars and leaf edge chewing by free feeding species were 12.2Â and 1.3Â greater on tied than on non-tied leaves, respectively. July and September leaf phenolic content had a slight positive effect on the probability of skeletonization, none on the probability of edge damage, and a weakly positive or negative effect on the intensity of skeletonization and edge damage, depending on leaf position.3. Second, we created experimental leaf ties, protected from herbivores, on the same Quercus species to determine whether tie formation changes leaf quality (total phenolics, nitrogen, water, toughness). Tie formation decreased phenolics, but this change was predicted to add only 0.8% leaf area loss. 4. Synthesis. Herbivory increased dramatically when leaves were in ties, with the effect mostly due to the tie itself rather than a change in leaf quality. We predict that the advantages of building and using leaf ties in this system are more likely to be escape from natural enemies and changes in abiotic factors.